Representations Of Morphemes Research Articles

From decomposition to distributed theories of morphological processing in reading.

The morphological structure of complex words impacts how they are processed during visual word recognition. This impact varies over the course of reading acquisition and for different languages and writing systems. Many theories of morphological processing rely on a decomposition mechanism, in which words are decomposed into explicit representations of their constituent morphemes. In distributed accounts, in contrast, morphological sensitivity arises from the tuning of finer-grained representations to useful statistical regularities in the form-to-meaning mapping, without the need for explicit morpheme representations. In this theoretically guided review, we summarize research into the mechanisms of morphological processing, and discuss findings within the context of decomposition and distributed accounts. Although many findings fit within a decomposition model of morphological processing, we suggest that the full range of results is more naturally explained by a distributed approach, and discuss additional benefits of adopting this perspective.

Modelling Maltese noun plural classes without morphemes

ABSTRACT Word-based models of morphology propose that complex words are stored without reference to morphemes. One of the questions that arises is whether information about word forms alone is enough to determine a noun's number from its form. We take up this question by modelling the classification and production of the Maltese noun plural system, using models that do not assume morphemic representations. We use the Tilburg Memory-Based Learner, a computational implementation of exemplar theory and the Naive Discriminative Learner, an implementation of Word and Paradigm, for classification. Both models classify Maltese nouns well. In their current implementations, TiMBL and NDL cannot concatenate sequences of phones that result in word forms. We used two neural networks architectures (LSTM and GRU) to model the production of plurals. We conclude that the Maltese noun plural system can be modelled on the basis of whole words without morphemes, supporting word-based models of morphology.

How the brain composes morphemes into meaning.

Morphemes (e.g. [tune], [-ful], [-ly]) are the basic blocks with which complex meaning is built. Here, I explore the critical role that morpho-syntactic rules play in forming the meaning of morphologically complex words, from two primary standpoints: (i) how semantically rich stem morphemes (e.g. explode, bake, post) combine with syntactic operators (e.g. -ion, -er, -age) to output a semantically predictable result; (ii) how this process can be understood in terms of mathematical operations, easily allowing the brain to generate representations of novel morphemes and comprehend novel words. With these ideas in mind, I offer a model of morphological processing that incorporates semantic and morpho-syntactic operations in service to meaning composition, and discuss how such a model could be implemented in the human brain. This article is part of the theme issue 'Towards mechanistic models of meaning composition'.

Transition-Based Korean Dependency Parsing Using Hybrid Word Representations of Syllables and Morphemes with LSTMs

Recently, neural approaches for transition-based dependency parsing have become one of the state-of-the art methods for performing dependency parsing tasks in many languages. In neural transition-based parsing, a parser state representation is first computed from the configuration of a stack and a buffer, which is then fed into a feed-forward neural network model that predicts the next transition action. Given that words are basic elements of a stack and buffer, a parser state representation is considerably affected by how a word representation is defined. In particular, word representation issues become more critical in morphologically rich languages such as Korean, as the set of potential words is not bound but introduce the second-order vocabulary complexity, called the phrase vocabulary complexity due to the agglutinative characteristics of the language. In this article, we propose a hybrid word representation that combines two compositional word representations, each of which is derived from representations of syllables and morphemes , respectively. Our underlying assumption for this hybrid word representation is that, because both syllables and morphemes are two common ways of decomposing Korean words, it is expected that their effects in inducing word representation are complementary to one another. Experimental results carried on Sejong and SPMRL 2014 datasets show that our proposed hybrid word representation leads to the state-of-the-art performance.

Root-letter priming in Maltese visual word recognition

Abstract We report on a visual masked priming experiment designed to explore the role of morphology in Maltese visual word recognition. In a lexical decision task, subjects were faster to judge Maltese words of Semitic origin that were primed by triconsonantal letter-strings corresponding to their root-morphemes. In contrast, they were no faster to judge Maltese words of non-Semitic origin that were primed by an equivalent, but non-morphemic, set of three consonant letters, suggesting that morphological overlap, rather than simple form overlap, drives this facilitatory effect. Maltese is unique among the Semitic languages for its orthography: Maltese alone uses the Latin alphabet and requires that all vowels are written, making such triconsonantal strings illegal non-words to which Maltese readers are never exposed, as opposed to other Semitic languages such as Hebrew in which triconsonantal strings often correspond to real words. Under a decomposition-based account of morphological processing, we interpret these results as suggesting that across reading experience Maltese readers have abstracted out and stored root-morphemes for Semitic-origin words lexically, such that these morphemic representations can be activated by exposure to root-letters in isolation and thus prime morphological derivatives.

Number of Meanings and Number of Senses: An ERP Study of Sublexical Ambiguities in Reading Chinese Disyllabic Compounds.

In English, an extensive body of work in both behavioral and neuropsychological domains has produced strong evidence that homonymy (words with many distinct meanings) and polysemy (many related senses) are represented, retrieved, and processed differently in the human brain. In Chinese, most words are compounds, and the constituent characters within a compound word can have different meanings and/or related senses on their own. Thus, in order to resolve lexical ambiguity in Chinese, one has to consider the composition of constituent characters, as well as how they contribute to whole word reading, known as “sublexical ambiguity.” This study investigates how two types of sublexical ambiguity affect Chinese word processing. The number of meanings (NOM) and the number of senses (NOS) corresponding to the first character of Chinese compounds were manipulated in a lexical decision task. The interactions between NOM and NOS were observed in both behavioral results and N400s, in which NOM disadvantage effect was found for words with few-senses only. On the other hand, the NOS facilitation effect was significant for words with multiple-meanings (NOM > 1) only. The sublexical ambiguity disadvantage suggested that semantically unrelated morphemes are represented as separate entries. For characters with multiple meanings, one orthographic form is associated with more than one morphemic representation. In contrast, the sublexical sense advantage supported the idea that semantically related senses that shared a morphological root are represented within a single entry. The more senses listed in a morphological root, the stronger representation will be formed. These results suggest that two types of sublexical ambiguities are represented and processed differently in Chinese word recognition models and also demonstrate that how they interact with each other in the mental lexicon.

From the Perspective of the Pervasiveness of Sound Correspondences: 語音對應與苗瑤語比較研究——從普遍對應的角度看

提要 普遍對應指對應的表現形式在所有參與比較的語言中都有分佈。放寬這一要求會導致晚期借用混入比較，乃至造成對語源關係的誤判。從這一角度出發，本文考察了原始苗瑤語重構所依賴的語音對應基礎。無論是在普遍對應的嚴格要求下，還是放寬要求，苗瑤語之間的關係語素都是高階比低階多，這可以進一步確認它們之間的親緣關係。應用同樣的比較程式，漢語和原始苗瑤語之間的關係語素也顯示出同樣的分佈，因此，詞階分析支持苗瑤語與漢語之間的親緣關係，而非接觸關係。不可釋原則的應用同樣確認了這一判斷。不可釋原則指無法用施借語言的音韻系統來解釋受借語言中關係語素的表現，就說明二者之間的關係假定為親緣關係比接觸關係更為合理。幾個不可釋的例子可以在漢-苗瑤語關係語素中找到，因此，苗瑤語應該仍然歸屬於漢藏語系。

Sensitivity to morphological composition in spoken word recognition: Evidence from grammatical and lexical identification tasks.

Access to morphological structure during lexical processing has been established across a number of languages; however, it remains unclear which constituents are held as mental representations in the lexicon. The present study examined the auditory recognition of different noun types across 2 experiments. The critical manipulations were morphological complexity and the presence of a verbal derivation or nominalizing suffix form. Results showed that nominalizations, such as "explosion," were harder to classify as a noun but easier to classify as a word when compared with monomorphemic words with similar actionlike semantics, such as "avalanche." These findings support the claim that listeners decompose morphologically complex words into their constituent units during processing. More specifically, the results suggest that people hold representations of base morphemes in the lexicon.

The fruitless effort of growing a fruitless tree: Early morpho-orthographic and morpho-semantic effects in sentence reading.

In this eye-tracking study, we investigated how semantics inform morphological analysis at the early stages of visual word identification in sentence reading. We exploited a feature of several derived Italian words, that is, that they can be read in a "morphologically transparent" way or in a "morphologically opaque" way according to the sentence context to which they belong. This way, each target word was embedded in a sentence eliciting either its transparent or opaque interpretation. We analyzed whether the effect of stem frequency changes according to whether the (very same) word is read as a genuine derivation (transparent context) versus as a pseudoderived word (opaque context). Analysis of the first fixation durations revealed a stem-word frequency effect in both opaque and transparent contexts, thus showing that stems were accessed whether or not they contributed to word meaning, that is, word decomposition is indeed blind to semantics. However, while the stem-word frequency effect was facilitatory in the transparent context, it was inhibitory in the opaque context, thus showing an early involvement of semantic representations. This pattern of data is revealed by words with short suffixes. These results indicate that derived and pseudoderived words are segmented into their constituent morphemes also in natural reading; however, this blind-to-semantics process activates morpheme representations that are semantically connoted.

Towards Regularisation: Morphological Spelling in Several Editions of the Kalender of Shepherdes

Abstract This corpus-based study focuses on the graphemic realisations of several derivational suffixes in thirteen editions of the Kalender of Shepherdes, an early modern almanac published between 1506 and 1656. Morphological spelling, that is, the consistent representation of particular morphemes, is considered to be one of the most important criteria in research on the orthographic standardisation in English. The analysis of the graphomorphemic information available in the documents under consideration indicates that particular printing houses applied different combinations of spelling rules with regard to the variants of suffixes and were characterised by varying levels of consistency in the use of these graphemic representations. The new spelling variants of the suffixes were adopted partly as the printers’ own regularisation policy, and partly under the influence of normative writings.

Seeing stems everywhere: Position-independent identification of stem morphemes.

There is broad consensus that printed complex words are identified on the basis of their constituent morphemes. This fact raises the issue of how the word identification system codes for morpheme position, hence allowing it to distinguish between words like overhang and hangover, and to recognize that preheat is a word, whereas heatpre is not. Recent data have shown that suffixes are identified as morphemes only when they occur at the end of letter strings (Crepaldi, Rastle, & Davis, 2010, "Morphemes in Their Place: Evidence for Position-Specific Identification of Suffixes," Memory & Cognition, 38, 312-321), which supports the general proposal that the word identification system is sensitive to morpheme positional constraints. This proposal leads to the prediction that the identification of free stems should occur in a position-independent fashion, given that free stems can occur anywhere within complex words (e.g., overdress and dresser). In Experiment 1, we show that the rejection time of transposed-constituent pseudocompounds (e.g., moonhoney) is longer than that of matched control nonwords (e.g., moonbasin), suggesting that honey and moon are identified within moonhoney, and that these morpheme representations activate the representation for the word honeymoon. In Experiments 2 and 3, we demonstrate that the masked presentation of transposed-constituent pseudocompounds (e.g., moonhoney) facilitates the identification of compound words (honeymoon). In contrast, monomorphemic control pairs do not produce a similar pattern (i.e., rickmave did not prime maverick), indicating that the effect for moonhoney pairs is genuinely morphological in nature. These results demonstrate that stem representations differ from affix representations in terms of their positional constraints, providing a challenge to all existing theories of morphological processing.

The interface between morphology and phonology: Exploring a morpho-phonological deficit in spoken production

Morphological and phonological processes are tightly interrelated in spoken production. During processing, morphological processes must combine the phonological content of individual morphemes to produce a phonological representation that is suitable for driving phonological processing. Further, morpheme assembly frequently causes changes in a word’s phonological well-formedness that must be addressed by the phonology. We report the case of an aphasic individual (WRG) who exhibits an impairment at the morpho-phonological interface. WRG was tested on his ability to produce phonologically complex sequences (specifically, coda clusters of varying sonority) in heteromorphemic and tautomorphemic environments. WRG made phonological errors that reduced coda sonority complexity in multimorphemic words (e.g., passed→[pæstɪd]) but not in monomorphemic words (e.g., past). WRG also made similar insertion errors to repair stress clash in multimorphemic environments, confirming his sensitivity to cross-morpheme well-formedness. We propose that this pattern of performance is the result of an intact phonological grammar acting over the phonological content of morphemic representations that were weakly joined because of brain damage. WRG may constitute the first case of a morpho-phonological impairment—these results suggest that the processes that combine morphemes constitute a crucial component of morpho-phonological processing.

Templates as affixation of segment-sized units: the case of Southern Sierra Miwok

<p>In this paper, I argue that certain templatic effects in Southern Sierra Miwok (SSM) follow from affixation of moras and underspecified segments. Such an analysis avoids the assumptions of syllabified X-Slots in the representation of morphemes that previous analyses for SSM argue for (Sloan 1991). In contrast, my optimalitytheoretic analysis predicts the templatic restrictions over whole strings of segments through the affixation of segment-sized phonological elements.</p>

Auditory representations and phonological illusions: A linguist’s perspective on the neuropsychological bases of speech perception

This paper argues that speech perception includes grammatical—in particular phonological—computations implemented by an analysis-by-synthesis component (Halle & Stevens, 1962) which analyzes linguistic material by synthesizing it anew. Analysis-by-synthesis, however, is not always required in perception but only when the listener wants to be certain that the words or morphemes identified in the input signal correspond to those intended by the speaker who produced the signal (= parity requirements, see Liberman, 1996, Liberman and Whalen, 2000). As we will see, in some situation analysis-by-synthesis may generate ‘phonological’ illusions. A central assumption is that the representations of words or morphemes in perception involve distinctive features and are formally structured into syllables. Two perceptual modes are needed: phonetic and phonemic perception. In phonemic perception only contrastive aspects of sounds, i.e., the aspect of sounds associated with meaning differences, are searched for. In phonetic perception both contrastive and noncontrastive aspects of sounds are identified. The phenomenon of phonological ‘deafening’ will be shown to follow from phonemic perception.

Derivational morphology and base morpheme frequency

Morpheme frequency effects for derived words (e.g. an influence of the frequency of the base “dark” on responses to “darkness”) have been interpreted as evidence of morphemic representation. However, it has been suggested that most derived words would not show these effects if family size (a type frequency count claimed to reflect semantic relationships between whole forms) were controlled. This study used visual lexical decision experiments with correlational designs to compare the influences of base morpheme frequency and family size on response times to derived words in English and to test for interactions of these variables with suffix productivity. Multiple regression showed that base morpheme frequency and family size were independent predictors of response times to derived words. Base morpheme frequency facilitated responses but only to productively suffixed derived words, whereas family size facilitated responses irrespective of productivity. This suggests that base morpheme frequency effects are independent of morpheme family size, depend on suffix productivity and indicate that productively suffixed words are represented as morphemes.

Multiple dimensions of relatedness among words: Conjoint effects of form and meaning in word recognition.

Words can be similar with respect to form (viz., spelling, pronunciation), meaning, or both form and meaning. In three lexical decision experiments (48 ms forward masked, 116 ms, and 250 ms SOAs), targets (e.g., FLOAT) followed prime words related by form only (e.g., COAT), meaning only (e.g., SWIM), or form and meaning (e.g., BOAT). BOAT-FLOAT and SWIM-FLOAT type pairs showed reduced target decision latencies relative to unrelated controls when primes were unmasked, but not when they were masked, and the magnitude of facilitation increased with increasing prime duration. By contrast, COAT-FLOAT type pairs produced significant inhibition at the shorter two prime durations. In all three experiments, including at the shortest SOA, (BOAT-FLOAT) pairs that shared form and meaning differed from COAT-FLOAT type pairs that shared only form. We discuss the similarity of the BOAT-FLOAT pattern to that of morphological facilitation and argue that if the same mechanism underlies both outcomes then activation of a shared morphemic representation need not underlie morphological facilitation.

The Distribution of Laryngeal Segments in Capanahua

The aim of this study is to enhance the documentation of laryngeal segments in Capanahua and to contribute to studies of the interaction between prosodic structure and segments. Laryngeal segments in Capanahua show a number of positional effects. They can only surface within morpheme‐initial syllables. This study shows that within that position, the distinction between head and nonhead syllables plays a major role in governing the alternations shown by glottal stops. Coda glottal stops can only surface in head‐syllables since they inhibit glottal coalescence, a process triggered elsewhere. In contrast, in nonhead syllables, glottal stops fuse with the preceding vowel, resulting in glottalized vowels. The underlying representations of morphemes can only contain one glottal feature; that is, either [constricted glottis] or [spread glottis] but not both. However, in Capanahua, this restriction does not hold at the level of phonetic representation since glottal stops and glottal fricatives co‐occur when the g...

With a letter-searched prime, boat primes float but swim and coat don’t: Further evidence for automatic semantic activation

Letter search (LS) on the prime typically eliminates semantic priming (swim-float) and orthographic/ phonological (O/P) priming (coat-float) but not morphological priming (marked-mark). However, LS on the prime does not reduce semantic priming for low-frequency targets (Tse & Neely, 2007). These findings suggest that semantic activation survives LS but decays during LS to a low level that can be detected only with sensitive measures, which are afforded by low-frequency targets and morphologically related primes and targets. In the present research, we show that LS on the prime results in 0 msec of semantic priming (e.g., swim-float) and 11 msec of O/P priming (e.g., coat-float), both of which are statistically null, whereas the LS semantic+O/P priming effect for primes and targets that do not share a morpheme (e.g., boat-float) is a robust 37 msec. Discussion focuses on the automaticity of semantic activation and whether morphological priming is mediated by (1) a morphemic representation that is separate from semantic representations or (2) activation combined from semantics and orthography/phonology.

The morpheme gender effect

In three experiments we explored the mental representation of morphologically complex words in French. Subjects were asked to perform a gender decision task on morphologically complex words that were of the same gender as their base or not. We found that gender decisions were made more slowly for morphologically complex words made from a base with an opposite gender compared to words for which the gender of the base matches that of the derived noun. Similar results were obtained for words that are pseudo-morphologically complex while no effect was observed for non-morphological embedded words. Our results suggest that during gender identification of derived and pseudo-derived words, morpheme and pseudo-morpheme lexical representations are activated, as well as their gender information. Also our results suggest that the noun ending predictability effect observed in previous experiments could in fact reflect decomposition into morphemes.

Stammkonstanz ohne Stützformen

AbstractIn German as in several other languages, the written forms of stems are often more constant than their spoken forms. This has traditionally been described in at least two different ways. In one approach, spelling encodes the abstract sound representations of morphemes, therefore ignoring sound variation at successive levels; in another approach, the spelling of the stems is determined by privileged base forms. In this paper, I analyse the constant spelling of stems in inflection as a case of paradigm uniformity, doing without abstract sound representations and privileged base forms.